ESP32 PSRAM support

[WiFive]

It seem QPI very sensitive to impedance so I would say check your board, put esp-psram on your board and check it.

Too many delays, time to make batch without psram and strip some features. Wonder if "esp64" still "shipping this year".

[rudi ;-)]

It seem QPI very sensitive to impedance so I would say check your board, put esp-psram on your board and check it.

Too many delays, time to make batch without psram and strip some features. Wonder if "esp64" still "shipping this year".

i saw - the fw was just in time no times not download - is ok - no problem
i go forward in this theme allone

ok,
got some express parcels this weekend ( 1.8 flash, psram and other things )
and had a fast try on it from saturday over sunday morning

1. test mod with nano32 and 1.8V 128Mbit and 1.8V 32Mbit psram successful

but we have no shematic on nano32 ( Gravitech & MakerAsia )
so we can only think what is connected.

i used
PSRAM #CE -> GPIO16 and pulled up 10K VDD_SDIO
PSRAM CLK -> GPIO17
Datapins was shared with flash
MTDI ( GPIO15 ) was pulled up VDD3.3 for VDD_SDIO 1.8V from SoC regulator



next test

2. test mod with DevKitC and 1.8V 128Mbit and 1.8V 32Mbit psram successfull

and here we have on Hardware Resources -
- ESP32 Module Reference Design
- ESP32 Development Board Reference Design

i used the same again
PSRAM #CE -> GPIO16 and pulled up 10K VDD_SDIO
PSRAM CLK -> GPIO17
Datapins was shared with flash
MTDI ( GPIO15 ) was pulled up VDD3.3 for VDD_SDIO 1.8V from SoC regulator

and the log looks like this:

pimped Nano32 with 128Mbit Flash and 32Mbit psram :

nano32_128_32.png (59.02 KiB) Viewed 12111 times

pimped DevKitC with 128Mbit Flash and 32Mbit psram

Log_psram_DevKitC.png (60.16 KiB) Viewed 12111 times

never give up until you finish your work

sry for the bad soldering - but had no scope on this fast try .

DevKitC_Block.png (1.04 MiB) Viewed 12111 times

btw:
in few days comes Caregiver Board - then I test this on site, why the FW was not run.

then i go again in a weekend with this theme - and then comes longer silence time,
cause there are summer holidays - and other things .


esp64 comes after chin. new year 2018 for public
( no - i do not know really when )


nice sunday and nice week

best wishes

rudi

[WiFive]

Good! So the conspiracy end?

[rudi ;-)]

..
So the conspiracy end?

i think you have no DevKitC on the hand?
you want know the secret?

make this test:
take the DevKitC
de solder the Wroom Case shield that you see the SoC and Flash

connect the DevKitC
do a measuring on VDD_SDIO VCC ( flash pin 8 ) and GND
what you get?

now make the same with pull up directly MTDI bootstrap pin to VDD3.3
do a measuring on VDD_SDIO VCC ( flash pin 8 ) and GND
boot again
what you get?


so long espressif, jeroen and ivan makes silence on wrover shematic
so long we can not talk here open over this "mystic"
and there is no "conspiracy" WiFive - its fact.

SoC Time Stamp
392016
P6W255

and yes - today is sunday and we work on this
cause we want to know what here happens!
and what does esptool do ..... but this is an other "conspiracy"...
you have to check by self.

remember things on efuses and suggestion to do a "wire pullup"
this "secret" comes open spoken too at least very late.......

take the devKitC and make your own test on it like i mentioned.

best wishes
rudi


now i understand why the shematic is not public yet
and i think - there is then only a shematic for rev1 public later
think on my words -

[rudi ;-)]

the next think over:
--------------------------


why is it mentioned in the datasheet that we need to pull the bootstrap for VDD_SDIO value
but not mentioned that this wrover has an "extra" resistor?
...
...

some interna guys know exactly detailed why this all so is..
.. in few days we know this too.


test the steps on DevKitC like i mentioned.
make the measuring on VDD_SDIO
and you will make

[rudi ;-)]

you asked in the past
why they have done a new "batch" on SoC,
where the timestamp from SoC was on begin this year
and they knowed that the next batch comes..

think over why we have
112017 ( rev0 ) "magic key" in ROM
and
122017 ( rev1) "key magic" in ROM
..

i am angry cause there is no words on this from interna guys - i do not write "espressif" cause
not all worker are at espressif - there are own companies too as interna workers...
( i think ) - more know.

[rudi ;-)]

it is 4.667 kOhm
https://esp32.com/viewtopic.php?f=2&t=8 ... t=70#p9743

[rudi ;-)]

@WiFive
same here
forget my last 2-3 postings
the boostrap wire was done on wrong IO15 here too.

holy moly
( jeroen i hear you laughing - but you have right )

where is the next corner
i am so small now

[rudi ;-)]

the next think over:
--------------------------

......
test the steps on DevKitC like i mentioned.
make the measuring on VDD_SDIO
and you will make

yes rudi - we got all a big .. better next time don't think ...
take a rest now - it is better


sry guys

[ESP_Sprite]

Good to hear you found the problem. So you have the Lyontek chip working as well now?